Confocal Microscopy for Diagnosing Nodular Lesions
Confocal Microscopy for Diagnosing Nodular Lesions
RCM has been applied in the clinical arena for the diagnosis of nodular tumours such as NM, BCCs and others although these studies comprised a limited number of cases or heterogeneous populations (mix of flat, palpable and nodular lesions). Furthermore, few invasive SCCs have been subjected to RCM analysis so far.
The aim of our study was to analyse clinically nodular tumours of different origin. RCM imaging for diagnosing the lesions was possible in 86·4% (10·7% of cases showed a discordance between RCM and histological diagnoses) of all lesions and 6·4% of cases were not evaluable. Among the cases that were not evaluable by RCM, the major limitations of RCM were the presence of ulceration covering more than 50% of the lesion surface and the presence of significant hyperkeratosis. Although ulceration per se can be seen as a favourable factor as RCM can be applied directly on the dermis, in the real setting it generates a chaotic RCM image with backscattering due to fibrin, clotted blood and debris, without any possibility to distinguish the deeper portion of the lesion.
Considering RCM diagnostic accuracy, eight cases out of 140 (5·71%) showed a discordant diagnosis between histopathology and RCM: one DF characterized by giant macrophages simulating the RCM appearance of atypical melanocytes; one Spitz naevus was misclassified as NM; and one Seb K was diagnosed as SCC by means of RCM. The remaining misdiagnosed lesions belonged to rare benign skin tumours (i.e. angiomyxoma, apocrine hydrocystoma, xanthogranuloma) for which RCM diagnostic findings have been discovered, with the exception of xanthogranuloma, and this can explain well the discordance with the histopathological diagnosis.
In our study we tested the RCM diagnostic accuracy in analysing NM+Mets vs. all other nodular lesions and it reached 96·5% sensitivity and 94·1% specificity. Recently, Guitera et al. analysed 710 consecutive clinically equivocal lesions comprising BCCs, melanomas (thin and thick melanomas) and other lesions. The accuracy of the melanoma algorithm developed by the authors (based on seven independent features) was 94·5% sensitivity and 73·9% specificity. The lowest value of sensitivity and specificity of the Guitera study can be explained by the fact that most melanomas that were misclassified by RCM were thin melanomas presenting few RCM criteria, or naevoid melanoma, which represents a 'special' and challenging variant.
Although a larger number of NM and Mets have to be analysed, our study confirms that RCM is a good tool in diagnosing NM with the exception of fully ulcerated NM that are not candidates for RCM imaging. Among the RCM independent features based on discriminant analysis, the presence of cerebriform nests was a relevant criterion for NM+Mets diagnosis, consistent with previous data based on few cases. In different study settings 'cerebriform nests' showed a variable diagnostic utility, and always showed a good specificity but poor sensitivity. This could be related to the different proportion of NM included in the different studies. Conversely, in this study, where only nodular lesions were included, we demonstrated that this feature is a relevant criterion for the diagnosis of NM+Mets, showing high sensitivity and specificity.
Concerning BCC diagnosis, RCM was confirmed to be an excellent tool for this tumour, reaching 100% sensitivity and 99·4% specificity. These data are in line with previously published papers even if nodular BCCs seem to be more easily diagnosed than flat ones, and thus, RCM showed higher diagnostic accuracy compared with studies that included either nodular or superficial BCCs. It should be considered that nodular BCC represents an easy clinical and dermoscopic diagnosis in the majority of cases. However, RCM should be considered in those cases where diagnostic confidence is lower, and NM cannot be ruled out.
Interestingly, all SCCs but one were correctly classified by means of RCM. Although the number of cases was limited (n = 6), the presence of epidermal disarray and the absence of pagetoid spread and dermal nesting were highly suggestive for SCC diagnosis. Further studies are needed to confirm these data and to test if RCM can be applied in difficult-to-diagnose SCCs and in the presence of hyperkeratosis. Undoubtedly, more SCCs need to be subjected to RCM imaging to better define the limits and the diagnostic capability of RCM for this tumour.
Concerning the other diagnostic categories, RCM was not able to analyse the two pyogenic granulomas that showed a large area of ulceration. Recently, Moscarella et al. reported the advantage of RCM in the diagnosis of red nodules including pyogenic granulomas. The authors identified vascular lagoons and inflammatory infiltrate suggestive for the diagnosis. In both our pyogenic granuloma cases a large ulceration, which is common in these tumours, hampered the effectiveness of RCM in detecting diagnostic features. Further studies are needed to obtain a precise characterization of pyogenic granuloma features and to evaluate RCM usefulness.
To sum up, our study highlighted that RCM is a powerful tool in the diagnosis of nodular lesions and that the major limitations are represented by ulceration and hyperkeratosis. This has clinical implications because fully ulcerated and mostly hyperkeratotic lesions cannot represent an indication for RCM. Moreover, in the few cases where the tumour was located too deep to be detected by RCM, an apparently normal epidermis, junction and superficial dermis was observable with RCM. Thus, when faced with a nodular lesion, it should be borne in mind that a diagnosis could be considered only in cases where characteristic features are identified, whereas absence of criteria should lead to prompt excision because a deep malignant tumour cannot be ruled out.
In our study we selected nodular lesions based only on clinical presentation and thus, when selecting difficult-to-diagnose nodules, the sensitivity and specificity of RCM may vary. However, RCM can be considered a valuable tool for nodules because, in many instances, a thinned epidermis allowed a good visualization of 'informative' diagnostic structures located in the superficial dermis. Prospective multicentre studies including not just nodular lesions with a RCM diagnosis rendered before histopathological assessment will be needed to highlight further the light and dark of in vivo confocal microscopy for the diagnosis of skin tumours.
Discussion
RCM has been applied in the clinical arena for the diagnosis of nodular tumours such as NM, BCCs and others although these studies comprised a limited number of cases or heterogeneous populations (mix of flat, palpable and nodular lesions). Furthermore, few invasive SCCs have been subjected to RCM analysis so far.
The aim of our study was to analyse clinically nodular tumours of different origin. RCM imaging for diagnosing the lesions was possible in 86·4% (10·7% of cases showed a discordance between RCM and histological diagnoses) of all lesions and 6·4% of cases were not evaluable. Among the cases that were not evaluable by RCM, the major limitations of RCM were the presence of ulceration covering more than 50% of the lesion surface and the presence of significant hyperkeratosis. Although ulceration per se can be seen as a favourable factor as RCM can be applied directly on the dermis, in the real setting it generates a chaotic RCM image with backscattering due to fibrin, clotted blood and debris, without any possibility to distinguish the deeper portion of the lesion.
Considering RCM diagnostic accuracy, eight cases out of 140 (5·71%) showed a discordant diagnosis between histopathology and RCM: one DF characterized by giant macrophages simulating the RCM appearance of atypical melanocytes; one Spitz naevus was misclassified as NM; and one Seb K was diagnosed as SCC by means of RCM. The remaining misdiagnosed lesions belonged to rare benign skin tumours (i.e. angiomyxoma, apocrine hydrocystoma, xanthogranuloma) for which RCM diagnostic findings have been discovered, with the exception of xanthogranuloma, and this can explain well the discordance with the histopathological diagnosis.
In our study we tested the RCM diagnostic accuracy in analysing NM+Mets vs. all other nodular lesions and it reached 96·5% sensitivity and 94·1% specificity. Recently, Guitera et al. analysed 710 consecutive clinically equivocal lesions comprising BCCs, melanomas (thin and thick melanomas) and other lesions. The accuracy of the melanoma algorithm developed by the authors (based on seven independent features) was 94·5% sensitivity and 73·9% specificity. The lowest value of sensitivity and specificity of the Guitera study can be explained by the fact that most melanomas that were misclassified by RCM were thin melanomas presenting few RCM criteria, or naevoid melanoma, which represents a 'special' and challenging variant.
Although a larger number of NM and Mets have to be analysed, our study confirms that RCM is a good tool in diagnosing NM with the exception of fully ulcerated NM that are not candidates for RCM imaging. Among the RCM independent features based on discriminant analysis, the presence of cerebriform nests was a relevant criterion for NM+Mets diagnosis, consistent with previous data based on few cases. In different study settings 'cerebriform nests' showed a variable diagnostic utility, and always showed a good specificity but poor sensitivity. This could be related to the different proportion of NM included in the different studies. Conversely, in this study, where only nodular lesions were included, we demonstrated that this feature is a relevant criterion for the diagnosis of NM+Mets, showing high sensitivity and specificity.
Concerning BCC diagnosis, RCM was confirmed to be an excellent tool for this tumour, reaching 100% sensitivity and 99·4% specificity. These data are in line with previously published papers even if nodular BCCs seem to be more easily diagnosed than flat ones, and thus, RCM showed higher diagnostic accuracy compared with studies that included either nodular or superficial BCCs. It should be considered that nodular BCC represents an easy clinical and dermoscopic diagnosis in the majority of cases. However, RCM should be considered in those cases where diagnostic confidence is lower, and NM cannot be ruled out.
Interestingly, all SCCs but one were correctly classified by means of RCM. Although the number of cases was limited (n = 6), the presence of epidermal disarray and the absence of pagetoid spread and dermal nesting were highly suggestive for SCC diagnosis. Further studies are needed to confirm these data and to test if RCM can be applied in difficult-to-diagnose SCCs and in the presence of hyperkeratosis. Undoubtedly, more SCCs need to be subjected to RCM imaging to better define the limits and the diagnostic capability of RCM for this tumour.
Concerning the other diagnostic categories, RCM was not able to analyse the two pyogenic granulomas that showed a large area of ulceration. Recently, Moscarella et al. reported the advantage of RCM in the diagnosis of red nodules including pyogenic granulomas. The authors identified vascular lagoons and inflammatory infiltrate suggestive for the diagnosis. In both our pyogenic granuloma cases a large ulceration, which is common in these tumours, hampered the effectiveness of RCM in detecting diagnostic features. Further studies are needed to obtain a precise characterization of pyogenic granuloma features and to evaluate RCM usefulness.
To sum up, our study highlighted that RCM is a powerful tool in the diagnosis of nodular lesions and that the major limitations are represented by ulceration and hyperkeratosis. This has clinical implications because fully ulcerated and mostly hyperkeratotic lesions cannot represent an indication for RCM. Moreover, in the few cases where the tumour was located too deep to be detected by RCM, an apparently normal epidermis, junction and superficial dermis was observable with RCM. Thus, when faced with a nodular lesion, it should be borne in mind that a diagnosis could be considered only in cases where characteristic features are identified, whereas absence of criteria should lead to prompt excision because a deep malignant tumour cannot be ruled out.
In our study we selected nodular lesions based only on clinical presentation and thus, when selecting difficult-to-diagnose nodules, the sensitivity and specificity of RCM may vary. However, RCM can be considered a valuable tool for nodules because, in many instances, a thinned epidermis allowed a good visualization of 'informative' diagnostic structures located in the superficial dermis. Prospective multicentre studies including not just nodular lesions with a RCM diagnosis rendered before histopathological assessment will be needed to highlight further the light and dark of in vivo confocal microscopy for the diagnosis of skin tumours.
Source...